It has various monikers and related concepts including precision medicine, stratified medicine, pharmacogenomics, pharmacogenetics and predictive medicine. But, whatever you call it, it means big changes in health care. Sociologists, ethicists, medics and others have all been confronting it in recent years. Economists have been relatively slow on the uptake, though some have begun thinking about it (see for example here, here, here)

Some of my current work involves evaluating predictive medicine in the form of a screening intervention, and we have a paper in the pipeline discussing some of the implications for cost-effectiveness analysis. This work is presenting a number of new challenges but is also highlighting some opportunities for the optimisation of health care.

Over the next few months I will be introducing and discussing some of the potential implications of personalised medicine for our discipline. These will include familiar topics in health economics and will probably fall under the following headings:

1. Demand for health and health care
2. Need
3. Supply of health care
4. Provider behaviour
5. Health insurance
6. Costs
7. Economic evaluation
8. Decision modelling
9. Population health
10. New market failures
11. Equity

These may merge, change or disappear as I progress, but I hope to cover as many angles as possible. Hopefully, with your feedback, we might be able to help guide future work in this area.

One of the most useful things about Quandl is its ability to provide that data directly into a usable format. You can even download any of the datasets straight into R; here, I will show you how.

Let’s look at total health spending as a proportion of GDP in the UK. We first find the dataset in Quandl (which is found here) and then click download where we have a number of options. In this case let’s opt for R.

We copy and paste the code into R

df<-read.csv('http://www.quandl.com/api/v1/datasets/WHO/20600_56.csv?&trim_start=1995-12-31&trim_end=2010-12-31&sort_order=desc', colClasses=c('Year'='Date'))

And then we plot

ggplot(aes(Year,Value),data=df)+theme_bw()+labs(x="Year",y="Healthcare spending as % of GDP")+geom_line()

Simple. There is also an R package available in CRAN that enables you to access data from Quandl without using the website and customising the data set (selecting variables and dates). I suspect that this will make finding appropriate data much easier in future.

In December 2012, an article was published in The European Journal of Health Economics with the title ‘Patients whose GP knows complementary medicine tend to have lower costs and live longer’. We spotted a number of shortcomings in the analysis and reporting, to which we felt a response was worthwhile. Subsequently the authors of the original piece, Professor Peter Kooreman and Dr Erik Baars, wrote a reply. In this blog post we summarise the debate and offer some concluding thoughts.

The study

The study employed a large dataset (n~150,000) from a Dutch health insurer. The objective of the study was “to explore the cost-effectiveness of CAM compared with conventional medicine”. The study sought to find out whether different levels of cost or mortality were observed depending on whether or not an individual’s general practitioner (GP) was trained in complementary and alternative medicine (CAM). The authors specifically looked at GPs trained in anthroposophy, homeopathy and acupuncture.

The authors implemented both a linear and log-linear regression model to estimate the cost differences associated with different types of CAM-training. Separate regressions were carried out for each type of CAM, for four different age groups and for five different cost categories. This gave a total of 120 different coefficients (2 (models) x 3 (CAM approaches) x 4 (age groups) x 5 (cost categories)) for the cost difference associated with CAM-training. Eighteen (15%) of these coefficients were negative (indicating positive findings attributable to CAM training) and statistically significant at the 5% level. Three (2.5%) coefficients showed a greater cost associated with CAM training.

For mortality effects, the authors implemented both a fixed effects logit and a fixed effects linear probability model (LPM). In this case the groups were split by sex and, again, by type of CAM-training; additionally an overall effect of CAM-training was included. This gave a total of 24 different coefficients for the mortality difference associated with CAM-training. Four (16.7%) of these were lower and statistically significant at the 5% level; all from the LPM.

The authors concluded that “patients whose GP has additional CAM training have 0–30% lower healthcare costs and mortality rates, depending on age groups and type of CAM”; adding that “since the differences are obtained while controlling for confounders… the lower costs and longer lives are unlikely to be related to differences in socioeconomic status.”

The study’s faults

A major problem with the study is one of selection. Selection is important in this study; there is selection of individuals who decide whether or not to register with CAM-trained GPs and selection of GPs who choose to pursue CAM. Patients that register with CAM-trained GPs may have different characteristics from those who do not, and exhibit different levels of cost and mortality as a result of these characteristics, rather than of CAM itself. The risk-adjustment the authors perform is the only way they deal with selection, and the set of risk-adjusters is very small; including only age, gender and postal code. The authors defend their position by citing a paper suggesting that selection bias might operate in the other direction. Neither we nor the authors can prove this one way or another. To thoroughly address selection, a larger set of risk-adjusters should be included and an approach such as propensity score matching would have been superior to the model adopted by the authors.

In reporting and reflecting upon their analyses, the authors do not recognise the problems associated with multiple testing. The authors appear to misunderstand the familywise error rate and the implications of this for the results that are currently shown as statistically significant. The authors should have accounted for this, using a method such as the Bonferroni correction.

The primary claims of the study are that patients with CAM-trained GPs had “0–30% lower costs” and “0–30% lower mortality rates”. These claims can be found throughout the original study, including the title, and in the authors’ subsequent dealings with the media. We believe that the first claim is a ‘cherry-picked’ finding; the second is simply false.

With regard to costs, as identified in the authors’ reply, the 30% relates specifically to patients “aged 75 and above with an anthroposophic GP-CAM”. But there are some coefficients that show a greater cost associated with CAM-trained GPs. Yet the paper’s title and publicity statements focus on this significant result alone. This is not an accurate reflection of the cost implications for patients in general, and highlighting this cherry-picked result is a misleading representation of the overall effects. A more appropriate way of reporting the results would have been to present the expected cost differences across the whole sample.

The analysis of mortality is simply incorrect. Mortality risk is bounded by 0,1 but the linear probability model is unbounded; making it inappropriate to model mortality data. The logit model is designed for binary outcomes, and when this is employed the significance of the mortality differences disappears or is less than 5%. But even the logit is inappropriate for these data because mortality is an infrequent event (around 3% of the sample died). A probit model would be preferable and we suspect that, had a probit been employed, no significant differences would be found. In short, the ‘significant’ effects that the authors identify are due to incorrect model specification.

In their responses, the authors retreated from their original emphasis on the significance of the mortality results saying that “our results do not show any evidence that patients of GP-CAMs have higher mortality rates”. We agree with this re-statement. Nevertheless, the title of the paper remains “Patients whose GP knows complementary medicine tend to … live longer”, which the authors now appear to admit is false.

Closing remarks

The study was available in its current form, as well as earlier versions, long before it was published in the EJHE. As a result, the study’s inaccurate claims have been repeated in a number of papers that cite the work in relation to herbal medicine and CAM in primary care. The publicity sounding these claims, and the authors’ conduct with the media, has been discussed elsewhere (English translation).

We believe that the original study and the response pieces might be used as a case study to aid teaching. To this end we have provided material to the Health Economics Education website. In addition, please do consider commenting below to develop the discussion – whatever your thoughts on the matter. Do you see other flaws in the study design? Or maybe you think some of our comments are unfounded? Are there better ways of studying important questions such as these?

The paper for discussion this month is a working paper published by the National Bureau of Economic Research. The authors are Janet Currie and W. Bentley MacLeod. The title of the paper is:

“Diagnosis and unnecessary procedure use: evidence from C-section”

Following the meeting, a transcript of the discussion can be downloaded here.

Links to the article

Direct: http://www.nber.org/papers/w18977

RePEc: http://ideas.repec.org/p/nbr/nberwo/18977.html

Other: tbc

Summary of the paper

In this paper the authors develop a model of diagnostic skill as an element of provider quality that is separate from a doctor’s skill in performing procedures. The model shows that higher surgical skill leads to higher use of surgical procedures across all patients, while better diagnostic skill results in fewer procedures for the low risk and more procedures for the high risk. When doctors face a dichotomous choice between an intensive and a non-intensive procedure they have a threshold level of patient condition; above which patients receive the intensive procedure and below which they receive the non-intensive procedure. The doctor’s threshold level is dependent on their surgical skill and the pecuniary benefit associated with carrying out the procedure. Greater diagnostic skill improves the precision of the doctor’s estimate of a patient’s condition and therefore improves the matching between patients and procedures; leading to better health outcomes. Taking the model to data on C-sections, the most common surgical procedure performed in the U.S., the authors show that improving diagnostic skills from the 25th to the 75th percentile of the observed distribution would reduce C-section rates by 11.7% among the low risk, and increase them by 4.6% among the high risk. Since there are many more low risk than high risk women, improving diagnosis would reduce overall C-section rates by about 5% of total births. Moreover, such an improvement in diagnostic skill would improve health outcomes for both high risk and low risk women, while improvements in surgical skill have the greatest impact on high risk women. The results are consistent with the hypothesis that efforts to improve diagnosis through methods such as checklists, computer assisted diagnosis, and collaborative decision making may improve patient outcomes.

Discussion points

• Are there other aspects of physician skill that could be estimated in this way?
• Is the characterisation of a doctor’s payoffs accurate?
• To what other procedures could the model be applied?
• To what extent could this model inform non-dichotomous physician decisions?
• What are the key policy implications of these findings?

Missed the meeting? Add your thoughts on the paper in the comments below.

But let’s be a bit more positive. The elusive `Super QALY’, as it has been described, should come eventually. And when it does, it’ll be great! One of the reasons, I propose here, is that it has the power to reconcile many of the disagreements that currently fuel (hamper?) debate in our field. Hence, the super QALY might just allow us to get on with fussing over minutia issues of economic evaluation.

Trade-offs

There are necessary trade-offs in decisions of resource allocation. These might be described as the ‘positive’ tensions economists deal with; they relate to decisions that must be made, regardless of our values. The equity–efficiency trade-off is the main one here. But there are others. For example, health care interventions have the dual aim of increasing both the quantity and quality of an individual’s life. The QALY attempts to address this. However, the way we value quality of life also incorporates considerations of length of life in so much as ‘death’ is used in the valuation of health states. This is problematic, as has been discussed. Economists haven’t really gotten round to disagreeing about this yet, but there’s plenty else on which we disagree.

Disagreements

These might be described as ‘normative’ tensions. They concern what different economists think should and should not be done; mainly relating to the process of valuing health states. There are welfarists and non-welfarists. There are those who support societal preferences, and those who support capturing patient experience. It should be clear to most that neither side in these debates is wrong. Most health economists acknowledge the value of capturing utility as well as the importance of capabilities. Most will attach some value to society’s preferences and some to those of the individual.

A super-QALY solution

It’s never been completely clear what the ‘extra’ in extra-welfarism (as currently practiced) actually consists. The super QALY will surely formalise this; it could involve some completely non-welfarist notions. The most common idea of the super QALY is one where the current health-related QALY is weighted based on some equity considerations. So, if this is where economic evaluation is heading, we’re likely to end up with an extra step of estimating the equity impact of an intervention. But, while most studies seem to suggest that this might just be an add-on process, I think it would require a realignment of the methods we already use.

Equity analysis

There’s no need for me to reiterate the importance of equity considerations. Plainly we (economists, the public) care about needs, capabilities, opportunities and equality. How we define the equity analysis is incidental. More important is that we get on with doing it and just see what happens. There are lots of measures we could use and different approaches we could take. For arguments sake (and because I quite like it), let’s say the equity analysis is characterised by a ‘minimum capabilities‘ approach. Something similar to Daniels’s normal opportunity range. People could have the normal opportunity range, have fewer opportunities or have more opportunities. We can argue later about where the threshold lies. People below the threshold could be said to be in ‘need’. Again, argue about this later. States could be defined using a capabilities measure; let’s just say the ICECAP-A for now (though I don’t much like it). Here in the world of health economics we like 0-1 scales, so the ICECAP-A could be valued based on these anchors. So, let’s say 1 is the minimum capabilities or normal opportunity range threshold. Zero equates to being dead. Values can drop below zero where opportunity sets represent a state worse that non-existence. For the equity analysis we are not interested in utility or satisfaction, so the valuation would not be by the individual. Values could be elicited from society, possibly. The valuation technique could be a person trade-off, maybe. Or we could let ethicists come up with weightings. This framework, surely, would satisfy the non-welfarists.

Health utility analysis

I see no reason why the estimation of health benefits cannot be utility-based. Utilitarian satisfaction is sufficient if non-welfarist concerns are incorporated in an equity analysis. Personally I believe that whether this is based on experiences or preferences is largely inconsequential and that, in terms of health, most of the differences demonstrated between the 2 are a function of the elicitation methods. Therefore, utility analysis would remain largely unchanged. However, the value of 0 would change. Zero currently represents either being dead or in a health state equivalent to being dead, despite these two things not being of equivalent value to a person. Under the new framework there is no need to incorporate death into the health utility analysis, as it is accounted for in the equity analysis. 0 should represent the worst health state imaginable. There would be no negative values.

Cost-effectiveness analysis

These 2 analyses would then be combined to form a relatively routine cost-effectiveness analysis to address the efficiency of the intervention. The QALY would be calculated in the usual way, but the ‘Q’ would become ‘super’ by being a function of the 2 different outcomes. Tentatively this could be done by multiplying the two values (alternative formulations could be defined by societal values or by ethicists, depending on your wont). Costings would be carried out in the usual manner and a super ICER could be calculated. Furthermore, the net benefit approach could be implemented in the usual way; possibly with separate willingness-to-pay values for each input to the super QALY (indeed, they may be willingness to pay values from different agents). The table below summarises how the approach might accommodate the various tensions in health economics.

All public policies could be subject to an equity analysis in the way set out above. It is in no way health-specific. Each policy field could then us this to weight their usual outcomes measures – preferably utility-based – to estimate the cost-effectiveness of their intervention. At this point the super QALY makes it onto daytime TV and health economists form a new unelected chamber at the Palace of Westminster.

No doubt this explicitly extra-welfarist approach to the super QALY raises more questions than it is currently able to answer, but we need to get on with trying stuff like this. The super QALY has proven elusive to date but, if we do make it, it may solve a lot of our problems. We may find ourselves having to invent new things to argue about.

In many countries, the allocation of health care is often more equal than other goods – it is ‘special’. Its ‘specialness’ can be seen since we would consider its distribution in isolation of other social goods to be morally significant. We would find it morally repugnant if access to health care was determined on the basis of income or assets while some inequality in income is not necessarily objectionable. Health care should therefore be treated differently from mere commodities, such as clothing or cars. Clearly then, equality is an important concern, but equality of what exactly?

Equality of opportunity

Norman Daniels argues that of central importance to health care is the maintenance of equality of opportunity.  Daniels asserts that health care protects the range of opportunities available to an individual – the way they can participate in social, political and economic life. He identifies this as a distinctly Rawlsian theory of justice as fairness. Importantly, he notes that this equality of opportunity is not based on happiness, welfare or utility. He considers this a strength and points out that disabled individuals often rank their welfare higher than do people imagining life with such a disability, or indeed someone with an acute illness. But, the disability may cause a loss to capabilities and opportunities that should be addressed regardless of welfare. This, he discusses, is a weakness of cost-utility analysis.

The equality of opportunity thesis may be subject to some objections. In contemporary society, gender and ethnicity still play a role in determining one’s opportunities. This then may provide an argument for providing gender reassignment surgery or skin colour alteration to those for whom there would be no medical benefit. Basing equality on welfare or utility may not be subject to the same objections since the effect of such a surgery both physically and in altering physical features important to personal identity may be significantly negative in terms of well-being.

Luck egalitarianism

One of the greatest debates in current political and economic discourse surrounding the distribution of health care resources is the importance of personal responsibility. A popular standpoint is one of luck egalitarianism (I have discussed this before). Health care should iron out the inequalities over which the individual has no personal control and beyond that the individual should be responsible for maintaining their own health. To see it from a different angle – if we had two individuals with the same health state the distribution of health care between them should be weighted by prudence. For example, if the driver and passenger of a car were admitted to hospital after a crash which may be considered the driver’s fault, even if it were just a momentary lapse in concentration, the passenger would have a greater claim to health care. However, in this situation, luck egalitarianism does admittedly seem too harsh. Supporters of this school of thought often argue that smokers, the obese, drug addicts and so forth have less of a right to health care, since they were aware of the risks of their actions but undertook them anyway.

I personally believe luck egalitarianism to not be an adequate account of justice. One’s physical reaction to heavy drinking or smoking is to a great extent determined by factors out of ones control, such as genes and socioeconomic factors. Pregnancy might be argued to have been a choice and so should not be supported under luck egalitarianism. Similarly, luck egalitarianism has difficulty distinguishing between reconstructive surgery and cosmetic surgery. An individual’s welfare may be affected by their appearance to some extent, something which they may have no control over, thus, providing cosmetic surgery would be supported.

The priority view

These previous accounts have all been of egalitarianism. However, egalitarianism faces an important objection, raised by Derek Parfit and others. The goal of egalitarianism in health care is to ensure an equality of opportunity or of utility, for example. However, this could easily be achieved by reducing the opportunities or utility of those at the top of the scale. This would certainly be rejected as a course of action. Parfit calls this the ‘leveling down’ objection. He revises egalitarianism and instead proposes prioritarianism or the ‘priority view’. Resources should be distributed in society weighted by where you are in the distribution – those at the bottom of the scale should receive greater benefits. This would reduce inequality while not being subject to the leveling down objection. In this situation, we could imagine a luck prioritarian position or modifying any of the other previously mentioned ideas.

England’s current system of allocation, as maintained by NICE, could be characterised as egalitarian. However, I might argue that it is only weakly egalitarian. It is not aiming to ensure everyone has the same level of utility; rather that everyone has the same opportunity to improve utility. In general, it does not take into account prudence or age or any other personal characteristics. This would have the effect of moving everyone’s health upward and would be egalitarian in the sense of reducing the gap between bottom and top, but this is only because there is a limit to the improvements healthcare can make (QALYs do not go higher than one). If there were no limit to health improvements our current system would not affect the distribution of health but shift everyone equally up the scale. I also believe that opportunity is also a concern as well as utility and since opportunity is correlated with health and quality of life, reducing inequality of one should reduce the inequality in the other. I think, then, that a prioritarian position is perhaps the most tenable – we should favour health care interventions that benefit the least healthy. What weights might be attached to the worst off is open to debate and the philosophical dilemmas to do with aggregating welfare still stand, but in any case, I think the priority view is better than our current system.

From health care to health

As a final note, I will say that I have only discussed the distribution of health care. More and more evidence is showing that as a determinant of overall health, health care is only a small contributor. Health care is ‘the ambulance waiting at the bottom of the cliff’. To extend the above theories to health rather than health care is problematic. We cannot redistribute health directly, so must redistribute the social determinants of health such as housing, income, autonomy in the workplace, etc. In this case, favouring a health distribution on the basis of ability to pay (favouring the poor) would not be morally repugnant. Does this mean the health is not a ‘special’ good, whereas health care is? It at least means that health should be treated differently to health care. In any case, evaluating these ethical and philosophical arguments can only strengthen the way we make these decisions. Perhaps ethics should be more widely taught to policy makers, economists, and others.

Read more

Arneson, R.J., 2000. Luck Egalitarianism and Prioritarianism. Ethics, 110(2), pp.339–349.

Daniels, N., 2001. Justice, health, and healthcare. The American journal of bioethics : AJOB, 1(2), pp.2–16.

Segall, S., 2010. Is Health (Really) Special? Health Policy between Rawlsian and Luck Egalitarian Justice. Journal of Applied Philosophy, 27(4), pp.344–358.

The paper for discussion this month is a working paper published in the Munich Personal RePEc Archive. The authors are Lydia Lawless, Rodolfo Nayga and Andreas Drichoutis. The title of the paper is:

“Time preference and health behaviour: A review”

Following the meeting, a transcript of the discussion can be downloaded here.

Links to the article

Direct: http://mpra.ub.uni-muenchen.de/45382

RePEc: http://ideas.repec.org/p/pra/mprapa/45382.html

Other: tbc

Summary of the paper

Time preferences affect individuals’ consumption decisions. Our understanding of time preferences can inform public policy, particularly in the area of health behaviours. Furthermore, in economic evaluation in health care, assumptions about time preferences play a crucial role in determining the cost-effectiveness of an intervention. The authors carry out a literature review; focussing on papers published post-2002 so as to avoid repeating previous reviews. In this review the authors sought to:

1. examine the influence of time preferences on health behaviours
2. explain how the societal time discount rate differs from the private time discount rate
3. determine how time discount rates affect the decisions of governments in the developing world
4. assess how time discount rates affect individuals’ decision making in regard to risky behaviours such as smoking, diet and sexual behaviour
5. discuss the repercussions of time preferences for the prevention of poor health.

The authors identified 3 main strategies that are used to capture time preferences; observed behaviour, experimental settings and the use of time preference proxies. The authors conclude that context plays a key role in determining the nature of time preferences; developing countries may exhibit different trends to developed countries. Furthermore, time preferences from a societal perspective do no necessarily match those of the individual.

Discussion points

• Do the authors succeed in reviewing all relevant literature?
• Is the authors’ review strategy sufficient?
• Does the study successfully address the 5 aims set out in the introduction?
• How might this study inform future research?

Missed the meeting? Add your thoughts on the paper in the comments below.

In the last post, I looked at a cross section of data from the 2011 census. I presented some correlations between the proportion of individuals who were unemployed and the proportion reporting bad health. I, and I am certainly not alone, may argue that myriad other factors could cause this observed relationship. I can’t prove or disprove any hypothesis in the space that this blog permits but I will add the following figure in support of the relationship. Here, I took data from both the 2001 and 2011 censuses for all lower super output areas (LSOAs; geographical areas of approximately 1,500 people) and looked at the relationship between the difference in the proportion unemployed and the difference in the proportion reporting bad health between 2001 and 2011:

Given the long lag between 2001 and 2011, the arguments from the previous post, that this represents changes to structural unemployment rather than short term cyclical unemployment, may still stand. But, for whatever reason, there is a correlation between unemployment and self-reported bad health.

I should mention that the questions about health differed between the two censuses from three options in 2001: ‘good health’, ‘fair health’, or ‘bad health’, compared to five options in 2011: ‘very good health’, ‘good health’, ‘fair health’, ‘bad health’, and ‘very bad health’. I have compared here the percentage reporting the 2001 option ‘bad health’ to the combined ‘bad health’ and ‘very bad health’ option. You may think this is an affront to good data analysis, so to allay your fears I have provided versions of the following two figures that use only 2011 data. You will see that they tell the same story.

The increase to poor health as a result of increased socioeconomic deprivation is costly for a number of reasons. Considering healthcare, direct costs such as hospital admissions for physical and mental health problems may increase, along with the accompanying costs of providing pharmaceuticals and other treatments. One cost that is not well reported in the media is that of unpaid care. One study in the UK estimated the costs of services provided by unpaid carers to be as much as £87 billion per year. Now, those in poor health require care. The following figure shows the relationship between the change in the proportion of people reporting bad health and the change in the proportion of people providing more than 20 hours a week of unpaid care between 2001 and 2011 in each LSOA:

2011 data only

I am not surprised by this relationship, and I doubt you are either. Then, it should also come as no surprise, given the previous two figures, that when I plot the relationship between the difference in the proportion unemployed and the difference in the proportion providing more than 20 hours unpaid care per week that there is also a strong relationship:

2011 data only

The relationship between health and economic conditions is complicated to say the least. What these data may indicate is that the cost due to increased unemployment may be far more than just reduced growth and output. Unpaid carers often have to leave employment to provide their services. Cutting back on health and social care funding in real terms will only shift the growing burden to individuals in poor areas, where health is worse, rather than to the state.

I would like to point out as a final note, and perhaps one of optimism, that the percentage of people reporting bad health has on average declined between 2001 and 2011. Although this may just be a case of hedonic adaptation…

1. Opportunity cost of time: In economic upturns leisure time decreases and health improving behaviours such as exercise decrease. Thus, in an economic downturn, since our time is less precious we have more time to engage in time-intensive and health-promoting activities. We could even visit the doctor more.
2. Health as an input to production: The production of goods and services requires healthy people. But this production may be hazardous or stress-inducing. Furthermore, some of the most hazardous sectors, such as construction, are the most affected by economic downturns.
3. External sources of death: Less time spent commuting means less time on the road and so fewer vehicular accidents. We may also see less drink driving, which is more common in economic upturns.
4. Income effect: Our consumption of alcohol and tobacco as well as other goods that damage our health may decline.

On the back of this evidence, I asked myself, has this effect been present in the UK during the current Great Recession? Overall, unemployment has risen over the last five years, and the average weekly wage has declined in real terms (thanks to @peterpannier for the graph):

A proper analysis of the data would be a full paper, something that someone, somewhere, may be in the process of writing – but, for the purposes of a preliminary investigation, let’s just look at the raw data. The 2011 census asked people how they would rate their health and provided them with five possible responses from ‘very good’ through to ‘very bad’. The census also provides us with the number of economically active but unemployed individuals. All this information is aggregated at the level of lower super output area (LSOA); of which there are around 32,000 in the UK each with a population of around 1,500. The following figure shows a plot of the proportion of unemployed individuals (as a proportion of 16-74 year olds) against the proportion reporting ‘bad’ or ‘very bad’ health:

Clearly, there is a strong upward trend; areas with more unemployed have more people reporting bad health. Does this contradict our initial hypothesis? One of the crucial points about the aforementioned arguments are that they are arguments to explain the relationship between a change in health and a change in economic circumstances. The papers cited above used a fixed effects analysis; an analysis to examine the effects of changes. Thus, the correlations in the figure above may be picking up structural unemployment: we may be seeing the relationship between health and unemployment for those for whom the recession doesn’t affect health behaviour because they don’t experience a change as they are already unemployed. So let’s look instead at the relationship between short-term unemployment and the proportion reporting ‘bad’ or ‘very bad’ health. I defined short term unemployed here as having last been employed in 2011, i.e. a maximum of three months prior to the census. I looked at this in two ways; firstly, by looking at the number of short term unemployed as a proportion of the total number of people between 16 and 74:

As you can see, there is now a downward trend, albeit not very steep. One issue is that areas with high short-term unemployment may also have high long-term unemployment making it hard to distinguish their effects. Therefore, my second approach was to look at the proportion of short term unemployed as a proportion of the total unemployed:

Now there is clearly a strong downward trend. At a superficial level, these data seem to preliminarily support the hypothesis that short-term changes to unemployment may improve health. However, we also see that long-term unemployment is related to negative health. This is certainly not unexpected.

It is well evidenced that longer spells of unemployment lead to a reduced probability of finding work. From the macroeconomic point of view, the longer a downturn in the economy lasts, the greater the structural unemployment. This, as the above data suggest, may therefore lead to a reduction in average population health. Reducing unemployment and the duration of employment spells is certainly important but an ambitious policy goal. A better understanding of how socioeconomic deprivation and poor health are related would identify other methods to combat this negative effect on health.

These data may also shine a different light on Keynes’s well quoted line that ‘In the long run we are all dead’.

Existing work

Plenty of work has been done on how to measure health inequality. It seems a pretty heinous crime to talk about health equality without mentioning Culyer and Wagstaff. More recently, new models of health inequality have been developed that bare varying levels of equivalence to a standard concentration curve (see here, here, here, here etc). But the authors of the aforementioned article are really interested in pure health inequality, irrelevant of income or socio-economic indicators. Some work has been done here too (see here, here, here etc); indeed, the age-at-death distribution thing was done by Le Grand.

Pure health inequality

Health and income are very different in a number of ways, and it seems a misnomer to compare income inequality with health inequality. The most important difference, probably, is how society views the two. Society has some aversion to income inequality and also aversion to health inequality. However, we don’t just prefer a more equal distribution of health; we want equal full health (i.e. health maximisation). Assuming diminishing marginal returns to health care (in terms of health), we will tend to prioritise those in worse health and tend towards equality. I would argue that health can only increase indefinitely in terms of longevity. We may live longer and longer but I think ‘full health’ is a very real ceiling while we’re alive. It simply isn’t possible for a super-rich elite to develop in terms of health. What would these people be like? Bionic presumably, but that’s a different debate. Even if health could be amassed indefinitely it wouldn’t be, as health has no value in exchange.

For me (given society’s aversion to inequality, technological progress and a maximum level of health at any point in time), movement towards equal health seems inevitable. You don’t need to agree with the Grossman model to accept that health represents a kind of ‘stock’. It therefore bares more resemblance to wealth than to income. Health requires some effort to maintain, but not to the same degree as income. Ho and Slavov’s article also introduces the idea of a lottery; luck plays an important role here. Society reacts differently to an income shock (say, unemployment) than it does to a health shock (say, being hit by a car). As with income there might be fair and unfair inequalities, but either way society is going to attach more weight to reimbursing an individual’s loss of health than an individual’s loss of income (unless, maybe, the latter is a result of the former). The same applies to those dealt a nasty hand at birth. In countries where health care is dependent on ability to pay there will certainly be more of a link between health and income; and thus between health inequality and income inequality. In countries like the UK, income inequality seems less likely to affect health inequality.

Health is becoming more equal; I won’t disagree with that. But, for the reasons outlined above, this seems somewhat inevitable. I suppose that doesn’t mean we shouldn’t celebrate it, but it does raise into question the value of doing so when there are real discrepancies between different demographics’ health that need addressing.

Cynics may spot the benefit of such an approach for those at the top of the income distribution…